Mapping (Interface Manipulation Package)

[using it]

Interface Manipulation Package v4.0 (2008-06-25)

©copyright

imp.classifier
Class Mapping

java.lang.Object
  imp.classifier.Mapping

public class Mapping
extends Object
extends Object

Defines a numerical data mapping.

Formally this may be defined as a function: double[dim] -> double[dim] which associated to a numerical value vector a low-dimensional representation of the data set.

Very easily a user construct a mapping and:

add a data of to this data set,
get the reduce representation of the data set,

as defined by manifold learning methods. The present class provides an interface for manifold learning mechanisms and implements local geodesic distance estimation.

Constructor Summary
`Mapping(double step, int size)` Constructs a mapping for the given parameter.

Method Summary
`Mapping`	`add(double[] data)` Add a data sample and return this.
`double[]`	`getData(int index)` Gets the n-th input data point.
`int`	`getDim()` Gets the reduce representation dimension.
`double`	`getDistance(int i1, int i2)` Gets the local geodesic distance between two data point.
`double`	`getDotprod(int i1, int i2)` Gets the local geodesic dot-product between two data point.
`double[]`	`getMap(int index)` Gets the reduce representation of a given input.
`int[]`	`getNeighbors(int index)` Gets the local neighborhood of the n-th input data point.
`int`	`size()` Returns the number of data input.

Methods inherited from class java.lang.Object
`equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait`

Constructor Detail

Mapping

public Mapping(double step,
               int size)

Constructs a mapping for the given parameter.

Parameters:: step - [optional, default is huge] Maximal distance between two points in the same neighborhood.; size - [optional, default is huge] Maximal size of a neighborhood. Either a size or a step must be defined.

Method Detail

add

public Mapping add(double[] data)

Add a data sample and return this.

Parameters:: data - data input value.

getData

public double[] getData(int index)

Gets the n-th input data point.

Parameters:: index - data index.
Returns:: The input data point.

getDim

public int getDim()

Gets the reduce representation dimension.

Manifold learning algorithms override this method.

getDistance

public double getDistance(int i1,
                          int i2)

Gets the local geodesic distance between two data point.

Parameters:: i1 - data point index.; i2 - data point index.
Returns:: The Euclidean distance between the data points if in the same neighborhood, else -1.

getDotprod

public double getDotprod(int i1,
                         int i2)

Gets the local geodesic dot-product between two data point.

Parameters:: i1 - data point index.; i2 - data point index.
Returns:: The Euclidean dotprod between the data points if in the same neighborhood, else 0;

getMap

public double[] getMap(int index)

Gets the reduce representation of a given input.

Manifold learning algorithms override this method.

Parameters:: index - data index.
Returns:: A low-dimensional representation of a data.

getNeighbors

public int[] getNeighbors(int index)

Gets the local neighborhood of the n-th input data point.

Parameters:: index - data index.
Returns:: An array of the index of the data points in the neighborhood, in distance increasing order.

size

public int size()

Returns the number of data input.

imp.classifier Class Mapping